Genetically engineered synthesis of natural products: from alkaloids to corrins.

INTRODUCTION

For almost 40 years organic chemists have been studying the metabolic pathways leading to the so-called secondary metabolites of nature. More than 100,000 of these small molecules (molecular weight [less than] 2500) have been characterized, and the biosynthetic origins of the major classes have been firmly established by tracer methodology using 14C- and 3H-precursors that were fed in vivo to plants and microorganisms. The next technical breakthrough involved cell-free systems and, where appropriate, the isolation and refeeding of intermediates accumulated in mutant bacterial cells. Slowly, by the end of the 1970s, the three main pathways to secondary metabolites were recognized. The origin of a majority of natural products could be assigned to the mevalonate, polyketide, or shikimate routes (and their combinations) by inspection of their structures and, in many instances, by experimental feeding or bioconversion studies using stable isotopes (13C, 2H, 15N) and NMR spectroscopy. By the early 1970s, however, it became clear that the complete definition of a pathway to a natural product would require the isolation and characterization of the full set of biosynthetic enzymes; often as many as 15 to 20 are required for the synthesis of a complex structure. Faced with the enormous task of identifying all of the intermediates and the enzymes that connect them in a multistep process, workers in the field began to focus on the genetic analysis of natural product biosynthesis using both genomic and cDNA libraries to search for the genes encoding the biosynthetic enzymes of the pathway. This review is concerned with recent advances accruing from experiments in which bioorganic chemists have changed their role from spectators of the natural synthesis of complex metabolites to active participants by attempting to reconstitute the pathway in vitro and (by using gene manipulation) to alter the structure of the final product, making full use of the combined power of organic chemistry and genetic engineering. The examples below are representative of this rapidly growing field and also are general enough in concept and execution to be applied to most plant, bacterial, or fungal systems.

The term "genetically engineered synthesis" must be defined in the context of the ensuing discussion in order to separate this endeavor from that of metabolic engineering of microorganisms. Great strides have been made in the latter field, especially in Streptomyces spp. in which genes may be mixed from several different organisms to produce hybrid structures corresponding to the combination of several pathways; progess has also led to amplification of selected genes from a metabolic pathway to over produce a target molecule. Metabolic engineering studies involve the in vivo production of desired or altered metabolites by an engineered strain. Similarly, transgenic plants can be engineered to produce enhanced yields of secondary metabolites or useful proteins by introducing pathways not present in the native species. All of the above techniques, which by now are securely ensconced in the repertoire of the fermentation and agrochemical industries, involve in vivo production with engineered cells or plants. In many cases, rigid branch points in the metabolic network can be removed, leading to higher yields of product (1). On the other hand, genetically engineered synthesis involves the preliminary acquisition of a complete set of soluble gene products followed by their recombination in vitro. This procedure eliminates unwanted processing of the key intermediates of the pathway along the main arteries of metabolism which, in vivo, is required for the well-being of the host organism. In other words, by removing all of the metabolic machinery except those enzyme-catalyzed reactions dedicated to the synthesis of the desired target, genetically engineered synthesis represents a new departure for the production of desirable and often rare natural products of biological importance.

The field is still in the early stages of development, and yet, from the examples described in this review, the successful acquisition of the biosynthetic enzymes appears to be the main hurdle. Earlier work with commercially available enzymes of the glycolytic pathway and recent work with overexpressed biosynthetic enzymes make abundantly clear that the combination of as many as 10 to 15 of these in a single reactor can lead to high overall yield of the target molecule and with perfect stereochemical fidelity.

COFACTOR REGENERATION

Many enzyme-catalyzed reactions require cofactors that are often too expensive to use stoichiometrically, especially on a large scale. Several efficient methods for the regeneration of cofactors have been developed that not only reduce the cost of the process but help drive the reaction toward product, either by influencing the equilibrium through coupling to the enzymes required for regeneration or by preventing the accumulation of by-products. Furthermore, the enantio-selectivity relative to the stoichiometric reaction is frequently increased when the catalytic cycle is coupled to a regeneration system. The principal cofactors and the methods presently available for their regeneration are discussed below.

Nucleoside Triphosphates (NTPs)

NTPs can be regenerated from their terminal reaction products, the di- or monophosphates using acetyl phosphate coupled with acetate kinase or phosphoenol pyruvate coupled with pyruvate kinase. Both systems have been used for large-scale syntheses of oligosaccharides in which the sugar nucleotide is regenerated (67).

An ingenious variant of this theme is the regeneration of ATP from ADP by harnessing the major segment of the glycolytic pathway. The glucose [right arrow] lactate process mediated by the 11 glycolytic enzymes has been coupled to the enzyme-catalyzed phosphorylation of creatine to achieve a one-flask conversion to phosphocreatine in 90% yield (61).

Nicotinamide Cofactors [NAD(P)(H)]

Several methods are now available for regeneration of these oxidoreductase cofactors. NADH and NADPH can be regenerated from NAD and NADP, respectively, with formate, glucose, or alcohol dehydrogenases. The reverse reaction, NAD(P)H [right arrow] NAD(P), is best accomplished with glutamate dehydrogenase or FMN reductase. When the equilibrium is favorable, the regeneration can be catalyzed by the same enzyme that is used for the synthetic step. The formate dehydrogenase-catalyzed conversion of NAD to NADH has been scaled up using a membrane reactor in which the NAD is covalently attached to polyethylene glycol to prevent loss through the membrane (62). An interesting addition to the repertoire of techniques for cofactor regeneration is the integration of cofactor dependent phosphorylation and dehydrogenation into a single closed loop, using phosphoenol pyruvate as a sacrificial reagent. This process allows not only the recovery of ATP and NAD+, but also the generation of dihydroxyacetone phosphate (DHAP), which can be further coupled with aldolases in a one-pot synthesis of keto sugars from glycerol and …